Colorful Physics for the Tevatron and LHC

Transcription

1 Colorful Physics for the Tevatron and LHC Andy Haas, SLAC SMU Video Seminar February 21, 2011 Slide 1

2 What is color flow?? QCD color charge is conserved locally It flows, just like electric charge Quarks carry one color (triplet) Gluons carry a color and an anti-color (octet) Others (W,Z,H,etc.) carry no color (singlet) Pulling apart a color from its anti-color takes a lot of energy (~1 GeV/fm) SINGLET color string or color connection formed Eventually color strings break by pulling quarks out of the vacuum hadronization OCTET Slide 2

3 How to see color flow Jet Pull Jets are not round shape is influenced by color flow!!! phi 0.5 RunII cone jet, loop over all cells in dr<0.7: phi Pull vectors point more towards each other for color singlet than octet θ where r points from jet center to each cell SINGLET eta Matt Schwartz and Jason Gallicchio Phys.Rev.Lett.105:022001, Slide 3 OCTET eta

4 How to see color flow Jet Pull Jets are not round shape is influenced by color flow!!! Pull vectors point more towards each other for color singlet than octet 0.5 RunII cone jet, loop over all cells in dr<0.7: θ where r points from jet center to each cell Matt Schwartz and Jason Gallicchio Phys.Rev.Lett.105:022001, Slide 4

5 Low mass SM Higgs Additional difference between H->bb and g->bb: H is color singlet b's must have same color and are connected together g is color octet b's have different color and are connected to beam Slide 5

6 Example from D0 Higgs analysis Signal Used for D0's ICHEP'10 ZH -> nunu bb search Extra ~5% sensitivity! ZH->nunubb (Could be improved to ~10%?) Background (mostly octet) shape well-modeled by MC But is the signal (singlet) shape well-modeled by MC? ( Color flow studied at LEP, but not jet pull, and these are hadron collisions...) Slide 6

7 Testing color singlet jet pull in data Need clean sample of W/Z/H decays to jets! ttbar is the most promising Have ~500 double b-tagged lepton+jet ttbar events at D0 ~90% pure sample W->jj decay is pure color singlet Verifying singlet color flow / jet pull for the first time ever at a hadron collider! Slide 7

8 Jet pull reconstruction Compute MC truth jet pull Use truth particles within dr<0.7 of jet and dz <1 cm of primary vertex (except muons and neutrinos) Do a Toy MC of calorimeter to see effects individually of: detector granularity energy thresholds / resolution detector noise / pileup Slide 8

9 Jet pull in control samples, data/mc Look at light jets in anti-b-tagged W+jets samples Good data/mc agreement, for both jets Leading-pT light jet Second-leading-pT light jet W + 2 light jets sample Slide 9

10 Separated by eta's of jets Minimum relative jet pull angle pt 1 pt 2 W + 2 light jets sample Slide 10

11 Checks of jet pull reconstruction Jet pull magnitude (amount of eccentricity) is also well-modeled But excess eccentricity is induced by the ICR Corrected for by adding a vector to the jet pull opposite to ICR direction Correction is also ~20% different for data/mc, used as systematic Different response/noise in ICR region ICR Pull biased towards +-eta direction ICR Data MC Slide 11

12 Checks of jet pull reconstruction Look at jets which were split and not split Noise/pileup area smaller towards other jet! Cells are assigned to the nearest jet Slide 12

13 OK back to ttbar! Added jet pull variables to standard D0 l+jets analysis 5.3/fb Extra Cuts: >=2 L4 b-tags 4 or >=5 jets mw-(jj) <30 Slide 13

14 Studies of ttbar lepton+jets 66% (46%) pure W->jj decays in M-mW <30 GeV for 4 (>=5) jets W->jj w -jets b-jets Slide 14

15 Studies of ttbar lepton+jets What do we expect from (full) MC? Separation is degraded by misidentified W->jj decays Slide 15

16 Jet pull in ttbar, data/mc Good data/mc agreement, for both w and b jets w-jets peak more towards zero (pointing towards each other)! But this is not an apples-to-apples comparison different kinematics and flavor for the light and b jets Next we do a direct comparison, in events with identical kinematics w -jets b-jets Slide 16

17 What color is the X in ttbar->bbw(->lnu)x(->qq')? Of course we know that X is a W, which is a color singlet... Test our sensitivity to separate singlet/octet and simulation modeling ttbar MC with OCTET hadronic X decay using MadGraph+Pythia ~10% separation Split into 5 regions based on dr, eta, and mw Central region, M-mW <30, with dr<2 is most sensitive (ttbar fit to singlet fraction observed in data) Data prefer a color singlet X Slide 17

18 Systematics on jet pull shape ALPGEN+Pythia agrees with ALPGEN+Herwig showering Light jets b-jets Slide 18

19 Systematics on jet pull shape MCNLO+Herwig agrees with plain Pythia Light jets b-jets Slide 19

20 Systematics on jet pull shape Various Pythia tunes, underlying event, other parameters, etc. All give similar jet pull shapes Tune ACR is flatter, unphysical color reconnection model for W Slide 20

21 Systematics on jet pull shape Recent Pythia tunes have a more physical way of turning on/off color reconnection, while still maintaining (most) agreement with UE No large effects from color reconnection (theoretically also <10%) Slide 21

22 Systematics Limited MC stats for singlet/octet templates Private production Managed to make ~1M events Still less than statistical uncertainty from data Next largest is data/mc correction for ICR region Slide 22

23 Systematics Other systematics (standard in ttbar analyses) are found to be very small... Slide 23

24 What color is the X in ttbar->bbw(->lnu)x(->qq')? Results! Submitted to Phys. Rev. Lett. Can't exclude octet W at 95% CL Data fluctuated by ~1 sigma? Or hint of imperfect simulations? Expected to exclude octet W at 3 sigma (99% CL) Slide 24

25 Color flow is extremely useful! Jet pull is a new, general tool that makes it possible to reliably use color information! Examples of what you can do with color flow: Learn new details of QCD (color reconnection, test string theory...) S Understand SUSY decay chains Can tell gluino decay from squark decay Help pair jets coming from the same decay Separate s-channel from t-channel single-top Technicolor (predicts some heavy octets) Many other BSM models MSSM Higgs (b(h->bb), h is singlet!) SM Higgs (W/Z(H->bb), tt(h->bb),...) O Slide 25

26 Summary Color flow is an interesting and useful tool, but only if you can simulate and reconstruct it reliably Jet pull is a new method which makes this possible Performed the first measurement of color flow in ttbar data Data contain a color singlet W->jj decay, within uncertainties You can sleep well tonight knowing that! Measures accuracy of simulations / jet pull for color-singlet decays Determines systematic uncertainty for jet pull in e.g. H->bb search With more data and studies, jet pull will become a precise tool Plan to look at jet pull in ATLAS simulations/data soon Slide 26

27 Backup Slides Slide 27

28 W matching fraction 4 jets >=5 jets Slide 28

29 Checks of jet pull reconstruction Look at jets which overlap and don't overlap for.5 and.7 cones W + 2 light jets sample Cells are assigned to the nearest jet Slide 29

30 Checks of jet pull reconstruction Bias of jet pull direction from non-uniformity of calorimeter vs. eta Particularly the ICR has lower response and more noise Fairly well-modeled by the MC, within ~20% Relative jet pull angle is not significantly affected W+2jet Data MC ICR +10% ICR +20% ICR +30% Jets touching ICR region Slide 30

31 Full MC vs. MC truth ~75% more separation of w vs. b pairs in MC truth Slide 31

32 Matching W in ttbar MC truth Slide 32

33 Jet pull in control samples, data/mc Look at light jets in anti-b-tagged W+3 jets sample Good data/mc agreement, for both jets Leading pt light jet Second-leading pt light jet Slide 33

34 Jet pull in control samples, data/mc Look at light jets in b-tagged W+3 jets sample Good data/mc agreement, for both jets Leading pt light jet Second-leading pt light jet Slide 34

35 Jet pull in control samples, data/mc Look at anti-b-tagged W+4 jets samples Good data/mc agreement, for both sets of jets Flatter than in real ttbar (since there's not real color connections?) Two jets with Mjj closest to mw Other two jets in the event leading-pt jet leading-pt jet Slide 35

36 Separated by eta's of jets W + 3 light jets sample Slide 36

37 Separated by eta's of jets W + 3 jets b-tagged sample Slide 37

38 Other eta,dr regions Slide 38

39 Other eta,dr regions Slide 39

40 Other eta,dr regions Slide 40

41 Other eta,dr regions Slide 41

42 Results in various sub-samples Slide 42

43 b-jets b-jets don't care much about the color representation of the X Slide 43

44 Pileup Amount of noise increases with #PV Changes average relative jet pull angles Separation between singlet and octet degrades only slightly Slide 44

45 Event display Slide 45

46 Event display Slide 46